EMPIRICAL ARTICLES an Exploratory Study of the Categorical

EMPIRICAL ARTICLES

An Exploratory Study of the Categorical Versus Spectrum Nature of Sexual Orientation

Ritch C. Savin-Williams Sex and Gender Lab, Department of Human Development, Cornell University

This exploratory study investigated the nature of sexual orientation (categorical or spectrum) by assessing the relative ability of sexual and romantic indicators to be predicted by sexual orientation labels. Young adults from a variety of community and college venues (N = 292) reported their sexual orientation label on a 9-point scale; from a 10-item list, their sexual identity; and the percentage of their sexual attraction, fantasy, genital contact, infatuation, and romantic relationship directed to males and females. Although the five indicators were significantly intercorrelated and sexual orientation labels predicted each indicator, discrepancies existed across indicators in relationship to sexual orientation (highest for attraction, lowest for romantic relationship). Sexual identity and sexual orientation label were strongly related at the ends of the sexual spectrum, less so in the middle. Men were nearly as nonexclusive as women. Study results supported the perspective that sexual orientation is a continuously distributed individual characteristic.

Sexual orientation is defined as an internal mechanism The spectrum perspective was noted nearly 30 years that directs a person’s sexual and romantic disposition ago by McConaghy (1987), who argued that sexuality toward females, males, or both, to varying degrees exists along a continuum with degrees of nonexclusivity (LeVay & Baldwin, 2012). It is manifested by a variety of in between heterosexuality and homosexuality. A decade indicators, including physiological arousal, erotic desire, later, in an invited critique of sex research, McConaghy sexual attraction, sexual fantasy, infatuation, genital (1999) concluded that this category versus continuum behavior, romantic relationship, and public and private debate regarding the nature of sexual orientation remained sexual identity (Sell, 1997). These indicators have been one of the major unresolved issues in sex science, reflected used to assess sexual orientation, either in terms of dis- in part by conflicting findings regarding whether sexual crete categories (heterosexual, bisexual, homosexual) or as orientation consists of taxa or is a matter of degrees existing along a continuum from exclusively to the oppo- (Gangestad, Bailey, & Martin, 2000; Haslam, 1997). site sex to exclusively to the same sex, with degrees of non- Although Kinsey (Kinsey, Pomeroy, & Martin, 1948) and exclusivity in between. Although investigators frequently others (e.g., Klein, Sepekoff, & Wolf, 1985) believed that theoretically accept and empirically assess sexual orienta- sexual orientation is continuous, Haslam (1997) noted tion as if it exists along this continuum (a 5– or 7–point that the Kinsey Scale, a 7-point scale to assess sexual ori- Kinsey-like scale), in practice they usually place research entation, more often than not has been used as a categori- participants into one of the three discrete, mutually exclu- cal measure. Because most people tend to congregate at sive groups. This dissolution of the sexual spectrum into the extreme ends of the Kinsey Scale, investigators use three categories is usually undertaken for methodological this pattern as justification to dichotomize research par- or practical considerations (e.g., small sample size in non- ticipants as either heterosexual or gay/lesbian; all extrane- heterosexual groups), although recently also for theoreti- ous (nonexclusive) orientations are labeled bisexual, cal reasons, especially for males (Bailey, 2009). although frequently grouped with gays/lesbians. The position assumed here is that dissolving sexual ori- The author is indebted to Dr. Gerulf Rieger for statistical assistance entation into three groups distorts our understanding of and comments on earlier drafts of this manuscript. sexuality. The prevalence of unacknowledged in-between Correspondence should be addressed to Ritch C. Savin-Williams, Director, Sex and Gender Lab, Department of Human Development, sexualities (between heterosexual and homosexual) can be Cornell University, Ithaca, NY 14853. E-mail: [email protected] substantial, sufficiently such that they should not be NATURE OF SEXUAL ORIENTATION grouped with another set or eliminated from consider- generalize to younger and older age groups was not ation. For example, research over the past decade has addressed in this study. Because male sexual orientation is revealed that mostly heterosexuals (Kinsey Scale 1) show a frequently construed as rigid, stable, and exclusive and unique physiological and behavior profile of sexual and female sexual orientation as flexible, fluid, and nonexclu- romantic characteristics that distinguishes them as a sepa- sive (Bailey, 2009; Baumeister, 2000; Diamond, 2008; rate sexual orientation group in between heterosexuals and Peplau, 2001), all analyses are presented here separately substantial bisexuals (Savin-Williams, Rieger, & Rosenthal, by sex. Finally, because of the recruitment strategy to 2013; Savin-Williams & Vrangalova, 2013). Indeed, there solicit a range of sexual orientations, the sample is not are more mostly heterosexual men and women than generalizable to a larger population that would be neces- bisexual, mostly homosexual, and gay/lesbian individuals sary to resolve the categorical versus spectrum dispute. combined (Savin-Williams, Joyner, & Rieger, 2012). A second example is reports of the behavioral and attrac- Method tion diversity within the bisexual group (Kinsey Scale 2 to 4) (Rodríguez Rust, 2002; Taywaditep & Stokes, 1998; Participants Weinberg, Williams, & Pryor, 1994; Weinrich & Klein, 2002), suggesting that the label bisexual itself covers mul- Of 325 individuals who responded to advertisements tiple subgroups. Third, in a nationally representative study about a study on gender and sexual orientation, 146 men of young adults there were nearly as many or more mostly (mean age = 21.7, SD = 3.4) and 146 women (mean homosexual (Kinsey Scale 5) individuals than male bisexu- age = 21.7, SD = 3.4) were 32 years or younger (90% of the als or lesbians (Chandra, Mosher, & Copen, 2011), with a total) and thus were included in analyses reported here. pattern of sexual attraction and sex partners distinct from Nearly two-thirds were currently in a university, commu- adjacent bisexuals and gays/lesbians (Vrangalova & nity college, or trade school (66% men, 61% women); the Savin-Williams, 2012). Finally, a significant minority of rest either had no further formal education after high heterosexual (Kinsey Scale 0) and homosexual (Kinsey school (10%, 9%), had graduated from college and were Scale 6) individuals report a small degree of attraction, currently working (11%, 16%), were enrolled in postgradu- fantasy, and/or behavior toward their less preferred sex ate work (7%, 8%), or had received a graduate degree (4%, and a majority of those with at least some same-sex attrac- 6%). Although the most common ethnicity was Caucasian tion do not identify as gay, lesbian, or bisexual (Ellis, (62% of men and 66% of women), the sample reflected a Robb, & Burke, 2005; Hoburg, Konik, Williams, & range of ethnicities, including mixed ethnicities (13%, Crawford, 2004; Morales Knight & Hope, 2012; Preciado, 11%), Asian (8%, 11%), African American (8%, 6%), Johnson, & Peplau, 2013; Savin-Williams et al., 2012; Hispanic (6%, 6%), and Native American Indian (3%, 0%). Vrangalova & Savin-Williams, 2010). The aims of this study were to provide data on whether Measures dimensional aspects of sexual orientation exist by explor- ing the varying indicators that have been identified by On a computer-based survey, participants completed a previous research. Although these sexual orientation 20-page questionnaire that included demographic infor- indicators are usually highly intercorrelated and are thus mation (age, sex, educational level, ethnicity) and other said to measure the same construct, this study explored measures used in this study. whether discrepancies exist within and across indicators such that individuals can be distributed not in a trifurcated Sexual orientation label. Sexual orientation label was system as straight, bisexual, or gay/lesbian but along a sex- assessed with an expanded version of the traditional ual orientation spectrum. The research questions follow: Kinsey Scale (i.e., two responses were added: exclusively straight and exclusively gay/lesbian). Participants responded to a prompt, “Please check the one that most RQ1: Which indicators (attraction, fantasy, sex behavior, infatuation, romantic relationship) have the accurately reflects your current understanding of your- strongest relationship to sexual orientation labels? self,” by checking one of nine responses: Exclusively RQ2: Is there a distribution of these indicators that sig- Straight = Only sexually attracted to the opposite sex; nificantly differentiates sexual orientation along a Straight = Nearly always sexually attracted to the oppo- continuum from exclusive heterosexuality to site sex. Rarely attracted to the same sex; Mostly exclusive homosexuality? Straight = Mostly sexually attracted to the opposite sex. RQ3: Is sexual identity related to sexual orientation? Occasionally attracted to the same sex; Bisexual-Leaning Straight = Primarily sexually attracted to the opposite sex Given considerable age differences in attitudes toward and and definitely attracted to same sex; Bisexual = More or the understanding and expression of sexuality (Jones & less equally sexually attracted to opposite sex and same Cox, 2010; Savin-Williams, 2005), the investigation was sex; Bisexual-Leaning Gay/Lesbian = Primarily sexually limited to the most recent, millennial, cohort—those ages attracted to the same sex and definitely attracted to the 18 to 32 years (Horovitz, 2012). Whether findings opposite sex; Mostly Gay/Lesbian = Mostly sexually

447

620.14 122.29 381.86 124.84 296.5 247.8 95.32 177.22 64.29 74.67 .97 .96 .88 .94 .88 .94 .85 .80 .91 .82 ======2 2 2 2 2 2 2 2 2 2 F F F F F F F F F F R R R R R R R R R R f f f e g g d d,e c,d c, d 10) 19) = = N N (87.0,113) (94.5, 102.7) (94.3, 103.6) (86.7, 106.1) (90.7, 106.1) (96.0, 102.4) (92.0, 104.8) (87.6, 112.4) (87.2,102.1) (87.3,102.0) f e e d d d f f g g 25) ( 28) ( = = N N (92.0, 99.7) (87.1, 99.3) (93.9, 99.2) (91.1, 99.2) (94.0, 100.8) (92.2, 104.4) (91.4, 101.1) (90.8, 103.1) (88.4, 105.6) (87.1, 103.1) f f f e c e c e d d,e 21) ( 17) ( = = N N (84.8, 91.6) (86.1, 94.8) (79.2, 88.0) (81.1, 90.9) (71.6, 88.9) (66.9, 80.3) (69.0, 87.3) (82.5, 93.1) (76.8, 94.7) (87.7,103.5) e e e c e c c d c,d d, e 10) ( 11) ( = = N N (68.0, 76.4) (74.6, 85.4) (75.1, 87.4) (60.5, 79.3) (65.0, 77.8) (51.2, 80.3) (46.7, 74.3) (57.1, 72.5) (71.4, 90.5) (54.6, 74.0) lower and upper 95% confidence levels; sexual orientation labels in row that share same letter share that in row labels orientation sexual and upper 95% confidence levels; lower = c d d d b d b d b c, d 7) ( 13) ( = = N N (55.2, 79.1) (48.2, 58.7) (53.8, 67.4) (40.9, 56.3) (18.2, 41.8) (49.5, 60.7) (13.7, 36.5) (51.1, 64.6) (11.1, 33.7) (47.3, 64.4) c c c c c c b a, b a, b a,b 9) ( 10) ( = = 32.6) , N N Straight Bisexual Gay Bisexual Mostly Gay Gay Gay Exclusive Statistics* Bisexual Bisexual (6.5 (35.9, 45.2) (34.5, 46.4) (24.0, 37.6) (27.5, 40.3) (34.8, 54.2) (33.9, 49.3) (14.9, 36.0) (6.0, 26.9) (6.7, 34.0) .0001. < a b b b b b a,b b p a,b a, b 19) ( 16) ( = = N N Mostly Straight (7.5, 17.7) (10.2, 17.2) (10.1, 19.1) (18.1, 27.4) (13.8, 27.8) (13.6, 24.7) (4.5, 20.4) (−4.1, 11.5) (1.7, 21.1) (−1.6, 17.4) a a a a a a a a a a 16) ( 20) ( = = N N (0.0, 6.2) (−0.8, 7.3) (−7.2, 7.7) (−2.6, 6.6) (−5.6, 8.7) (−2.1, 8.0) (−2.4, 3.0) (−10, 10.6) (−1.6, 10.6) (−9.4, 10.2) a a a a a a a a a a 22) ( 19) ( = = .05) differ from each other. from .05) differ N N 0 (0–0) 0.3 (0–5)0 (0–0) 11.4 (0–60) 0.4 (0–5) 20.3 (0–100) 7.9 (0–97) 25.1 (0–60) 19.6 (0–100) 65.8 (0–100) 22.4 (0–60) 78.2 (0–100) 97.0 (50–100) 60.5 (0–100)(100–100) 100 85.7 (10–100) 95.1 (25–100)(100–100) 100 ( ( Straight Straight 0.0 (0–0) 0.3 (0–3) 12.4 (0–50) 25.4 (0–71) 67.1 (20–100) 80.9 (20–100) 95.6 (81–100) 98.3 (80–100) 94.7 (4–100) 0.3 (0–5) 3.1 (0–20) 13.7 (5–33) 40.6 (30–65) 53.4 (36–81) 72.2 (50–83) 88.2 (70–99) 96.5 (80–100) 99.2 (90–100) < Exclusive Exclusive 0.8 (0–15) 1.6 (0–20) 3.7 (0–30) 16.4 (0–50) 30.0 (0–60) 69.9 (19–100) 80.2 (54–100) 96.9 (78–100) 94.7 (4–100) (−2.9, 3.5) (−6.5, 8.2) (−3.7, 4.6) (−7.7, 7.7) (−3.2, 6.2) (−3.4, 5.3) (−5.7, 7.4) (−4.2, 6.1) (−9.2, 9.2) (−8.5, 8.5) p mean percentage (range) of sexual indicator for each sexual orientation label. Second line label. orientation of (range) each sexual mean percentage for indicator sexual = + Down) Orientation Labels (Read Nine Sexual of Across Levels Orientation Indicators Sexual First line + . FantasyGenital Behavior 0.5 (0–5)InfatuationRomantic 3.3 (0–35) 1.5 (0–19)Relationship 2.0 (0–10) 14.6 (0–46)Attract 59.6 (50–90)Fantasy 12.6 (0–50) 60.6 (39–90)Genital 1.0 (0–5) 30.8 (9–51)Behavior 0.9 (0–10)Infatuation 80.0 (50–99) 48.6 (20–100) 2.9 (0–8) 5.3(10–25)Romantic 90.5 (80–100) 81.3 (60–100) 1.0 (0–10)Relationship 22.7 (9–55) 20.8 (10–83) 97.4 (75–100) 86.0 (70–100) 4.5 (0–29) 33.9 (10–71) 98.6 (90–100) 44.5 (15–70) 95.9 (70–100) 19.2 (2–50) 55.1 (20–80) 98.9 (90–100) 55.8 (29–77) 41.6 (20–70) 71.4 (50–95) 64.3 (0–95) 57.8 (40–91) 83.6 (61–95) 73.6 (40–95) 64.8 (5–100) 95.2 (75–100) 93.2 (50–100) 87.8 (70–100) 98.4 (90–100) 96.4 (90–100) 96.2 (80–100) 98.4 (93–100) Attract ANOVA for each indicator across sexual orientation labels at the at labels orientation sexual across each indicator for ANOVA do not significantly ( do not significantly Women Note Table 1. Table Label Indicator * Men

448 NATURE OF SEXUAL ORIENTATION attracted to the same sex. Occasionally attracted to the was subsequently conducted, and each of the five indica- opposite sex; Gay/Lesbian = Nearly always sexually tors was significantly predicted by sexual orientation label attracted to the same sex. Rarely attracted to the opposite for both sexes, with effect sizes (R-Squares) ranging from sex; or Exclusively Gay/Lesbian = Only sexually attracted .80 to .97 (Table 1). Sexual attraction most strongly differ- to the same sex. The distribution of men and women entiated mean levels across sexual orientation labels and it along these nine responses is presented in Table 1. produced the most (N = 7) nonoverlapping confidence intervals of group means (Table 1). Fantasy and infatua- Sexual orientation indicators. Participants indicated tion were nearly as effective; genital contact and romantic their current sexual attraction, sexual fantasy, genital con- relationship were slightly less effective. tact (“on the part of one or both of you”), infatuation/ Based on their overlapping 95% confidence intervals crush, and romantic relationship (“dating, both serious of means, all indicators failed to distinguish exclusively and not so serious”) directed to males and to females. straight from straight and exclusively gay/lesbian from gay/lesbian labels (Table 1). By contrast, for both sexes, Sexual identity. Embedded within the demographic sharp increases in same-sex sexuality occurred between section was the following question: “What is your sexual mostly straights and bisexual-leaning straights in attrac- identity (please check one)?” Responses included (in tion, fantasy, and infatuation, and between bisexuals and alphabetical order): Bicurious, Bisexual, Gay/lesbian, bisexual-leaning gays/lesbians in romantic relationship. + Mostly gay/lesbian, Mostly straight, Straight, Questioning, Among men, a dramatic increase ( 40%) in same-sex Queer, Unlabeled, and Other (please specify). genital contact occurred between bisexual-leaning straights and bisexuals; among women, between bisexuals and bisexual-leaning lesbians. Despite mean group Procedure differences that were frequently significant, each of the In 2010 a study on gender behavior and sexual nine sexual orientation labels overlapped (on all five indi- attraction was advertised in various locations in a rural, cators) in its range with at least one of its adjacent labels. college-town community. Advertisements promoting the Across indicators, R-Square values were slightly study were posted on several university Web sites and higher in men than women. A MANOVA indicated that electronic mailing lists. To enhance recruitment of non- the general relationship of orientation indicators with heterosexuals and noncollege students, advertisements orientation label was marginally stronger in men than were placed on a local Facebook page catering to sexual women, F (8, 250) = 1.90, p = .06. The MANOVA also minority populations; a Craigslist Web forum where men indicated that the sexes differed significantly in how sought both men and women for sexual reasons; and individual indicators related to sexual orientation labels, mailing lists for university sororities, fraternities, and resi- F (8, 250) = 6.27, p < .0001 (Table 1). This sex difference dence halls known for their open acceptance of nonhet- was most apparent in contrasting adjacent sexual orien- erosexual individuals. Potential volunteers were tation labels along the continuum from exclusive hetero- encouraged to share the lab’s e-mail address with others. sexuality to exclusive homosexuality (see next section). Participants contacted the lab by e-mail, and an appointment was arranged. The survey portion of the Comparisons of Adjacent Sexual Orientation Labels study was administered online using a Web surveyor tool (Qualtrics) in a confidential setting on a lab computer. Exclusively straights’ mean level of same-sex sexuality The survey assessed various aspects of sexual orienta- was zero only for genital contact (both sexes) and roman- tion, personality, and gender nonconformity and took tic relationship (women). On all indicators, straights were approximately 20 minutes to complete. Participants were slightly more same-sex oriented than exclusively straights compensated for their time and debriefed. The university’s and less so than mostly straights. It was among mostly Institutional Review Board for Human Participants straights (both sexes) in which same-sex sexuality had its approved the study. first considerable increase (except for romantic relationship). With the exception of genital contact, mostly straight women were more same-sex oriented than mostly Results straight men. Another relatively large increase in same-sex sexuality Sexual Orientation Indicators occurred among bisexual-leaning straights (both sexes). The five sexual orientation indicators were significantly In particular, bisexual-leaning straight men differed from intercorrelated for men (rs = .91 to .98, all ps < .0001) and bisexual men primarily in having substantially fewer women (rs = .81 to .96, all ps < .0001). A multiple analysis of same-sex genital contacts and romantic relationships. variance (MANOVA) indicated that the strength of these Whereas one-quarter of bisexual women reported sex relationships was significantly different among indicators, with a woman, two-thirds of bisexual men reported sex F (8, 125) = 8.75, p < .0001 for men and F (8, 125) = 6.85, with a man. Less than one-third of all bisexuals reported p < .0001 for women. An analysis of variance (ANOVA) a romantic relationship with someone of the same sex.

449 SAVIN-WILLIAMS

On all indicators, there was also a dramatic jump in data in Table 2). Agreement was most evident at the ends same-sex sexuality between bisexual men and bisexual- of the sexual orientation spectrum, with over 90% of leaning gay men. The increase in same-sex sexuality exclusively straights and exclusively gays/lesbians identi- was less pronounced between bisexual women and fying as straight or gay/lesbian, respectively. Although bisexual-leaning lesbians, except in genital contact and the majority of the middle five sexual orientation labels romantic relationship. For both sexes, on nearly all indi- and identities matched, a large minority of the nonexclu- cators, bisexual-leaning gays/lesbians were closer to sive labels incorporated a range of identities. For exam- mostly gays/lesbians than they were to bisexuals, and ple, mostly straights by sexual orientation label identified mostly gays/lesbians were closer to gays/lesbians than to as straight, bicurious, bisexual, queer, or unlabeled; and bisexual-leaning gays/lesbians. What most distinguished mostly gays/lesbians by sexual orientation identified as mostly gay men from gays was their significantly fewer bisexual, queer, gay/lesbian, unlabeled, or other. Neither same-sex crushes and romantic relationships; among sex was more likely to use nontraditional sexual identities mostly lesbians from lesbians, significantly fewer reports or to mismatch their sexual orientation and sexual of same-sex attraction, fantasy, and genital contact. identity. The distinction between gays/lesbians and exclusively The range of associated sexual identities and gays/lesbians mirrored findings from the heterosexual orientations was broad in several circumstances. end of the spectrum. The two groups generally did not Bicurious-identified individuals labeled themselves as differ on mean indicator scores, and only on genital con- mostly straight (the majority), bisexual-leaning tact and romantic relationship among women were the straight, or bisexual. Queer-identified individuals means 100% for exclusively gays/lesbians. ( primarily women) were any orientation other than exclusively straight or straight. The 10 unlabeled, questioning, or “other” individuals were also in the Sexual Identity middle of the continuum, populating every sexual ori- Sexual identity labels corresponded to sexual entation label except for the four straight and gay/lesbian orientation labels for both men and women (descriptive labels.

Table 2. Sexual Identities Among the Nine Sexual Orientation Labels

Exclusive Mostly Bisexual Bisexual Mostly Exclusive Label Identity Straight Straight Straight Straight Bisexual Gay Gay Gay Gay

Straight Men 95% 80% 6% Women 100% 88% 5% Mostly straight Men 20% 50% 11% Women 13% 74% 20% Bicurious Men 13% 11% 14% Women 21% 10% 15% Bisexual Men 5% 13% 67% 86% 55% 12% 4% Women 0% 0% 40% 46% 60% 10% 0% Mostly gay Men 18% 71% 29% Women 10% 76% 12% Gay/lesbian Men 6% 68% 100% Women 5% 68% 80% Queer Men 13% 11% 0% 18% 0% 0% 0% Women 0% 10% 23% 20% 5% 20% 20% Unlabeled Men 6% 0% 0% 9% 6% Women 10% 8% 10% 5% Questioning Men 0% Women 10% Other Men 0% 6% Women 8% 0%

450 NATURE OF SEXUAL ORIENTATION

Discussion be of any sexual orientation label except exclusively heterosexual or heterosexual. Despite the popular appeal Answering research question 1, each sexual orienta- and simplicity of these terms, participants were consis- tion indicator was significantly predicted by sexual orien- tent with the lack of cultural consensus regarding the tation labels. Given that the definition of the sexual appropriate sexual identity names that should be attached orientation label included the term sexual attraction, this to nonheterosexual and nonhomosexual individuals. was not surprising for the sexual attraction indicator. Supportive of this position, youth in one study reported Indeed, sexual attraction has been a common and reliable that a measure of sexual identity with the options hetero- method of determining sexual orientation, although it is sexual, bisexual, gay/lesbian, and unsure was the most dif- not necessarily clear to some research participants what is ficult to answer, largely because most preferred an meant by sexual attraction (Austin, Conron, Patel, & assessment tool that provided intermediate options which Freedner, 2007). As a single measure of sexual orienta- reflected their experience of feeling between categories tion, however, sexual attraction remains superior to oth- (Austin et al., 2007). Thus, to avoid ambiguity, research- ers, although this might depend on the specific research ers should use sexual orientation labels rather than sexual question (e.g., sexual behavior for HIV research) (Savin- identities. Williams, 2006). There were relatively few sex differences. Although Sexual orientation labels were the least good predic- correlations between sexual orientation labels and indi- tors (yet significant) of genital contact and romantic cators were slightly higher among men than women, relationship—possibly because engaging in sexual with women displaying marginally lower boundaries behavior and romance can be constrained by a number between sexual orientation labels, many men occupied of factors not directly related to sexual orientation (e.g., nonexclusive labels, counter to previous research (Bailey, availability of a willing partner, age, virginity pledge). 2009). In terms of sex differences, men experienced Haslam (1997) previously warned investigators to focus marked increases in same-sex genital contact and women less on overt sexual conduct to assess sexual orientation in same-sex romantic infatuation within the bisexual and to rely more on dispositional variables that are range (bisexual straight, bisexual, bisexual gay/lesbian). continuous, such as attraction and fantasy. In this study, Perhaps this reflects the greater significance of romantic however, genital contact and romance remained reason- indicators among women and genital contact among ably good indicators of sexual orientation. Whether men to switch from heterosexuality toward homosexual- millennial participants are more likely than previous ity. For both sexes, however, romantic relationship was generations to express their dispositional characteristics the one indicator that lagged most in becoming same-sex through sexual and romantic conduct is worthy of future oriented, with a marked increase occurring among investigation. bisexual-leaning gays/lesbians. To have a small degree of In regard to research question 2, there was a clear same-sex attraction and fantasy, to develop crushes on distribution of sexual orientation indicators along a spec- another girl or boy, or to engage in genital contact with trum with little evidence that participants coalesced into someone of the same sex and maintain some degree of two or three sexual orientation labels or identities. heterosexuality is perhaps possible for this millennial Depending on the indicator, they differentiated them- cohort, but to fall in love and have a romantic relation- selves into labels ranging in size from seven (attraction) to ship with another girl or boy likely reflects a more sub- four (romantic relationship). Although these labels were stantial personal and public marker of one’s same-sex statistically distinct, they did not match up on all indica- sexuality. tors. For example, bisexual women formed a unique An alternative possibility is that although sexual and group on attraction, but they did not on the other four romantic orientations are identical for most individuals, indicators (e.g., distinct from bisexual-leaning straights for some individuals the two operate somewhat indepen- and bisexual-leaning lesbians on fantasy). A second indi- dently of each other, especially in their timing of expres- cation of the noncategorical nature of sexual orientation sion. One study found that youth clearly distinguish was the considerable overlapping ranges within the sexual between their sexual orientation (a physiological, uncon- orientation labels across indicators. Thus, although sex- trollable desire for a specific gender, person, or attribute ual orientation labels had a characteristic mean level of of a person) and their romantic orientation (being or an indicator, each often shared its range with several wanting to be in love with a specific gender) (Friedman other adjacent labels. et al., 2004). Lags in the association between romantic In regard to research question 3, the use of sexual relationship and sexual orientation might confuse some identity to designate a sexual orientation label was gener- individuals as to the nature of their sexuality. Future ally accurate on the extreme ends of the continuum but investigations should assess separately sexual and roman- less so among nonexclusive individuals (the middle of the tic orientations, especially regarding intraindividual continuum). For example, bisexually identified individu- congruence. als could be of any sexual orientation label except hetero- One novel aspect of the study was the expansion of sexual or exclusively gay/lesbian; a queer individual could Kinsey’s original 7-point scale to nine by adding one point 451 SAVIN-WILLIAMS to each end of the continuum. This decision was based on Based on data presented here, heterosexual, bisexual, the considerable data that not all heterosexual and homo- and gay/lesbian labels do not constitute the universe of sexual individuals are exclusive to their preferred sex on sexual orientations. Although the usual procedure when all indicators (Vrangalova & Savin-Williams, 2012). In the sexual orientation indicators are highly intercorrelated present study, although separating exclusively heterosex- is to create a single composite score, this process dis- ual and gay/lesbian from merely being heterosexual or guises variability and complexity within the sexual ori- gay/lesbian did not result in distinctive groups, the latter entation construct. The results of this study indicate groups were slightly less exclusive than the former. Yet that sexual orientation is a continuously distributed despite the inclusion of the term exclusively, not all such characteristic of individuals, and all decisions to catego- identified individuals were so restricted. Exclusively rize it into discrete units, regardless of how many, are straight and lesbian women were only exclusive in genital ultimately external impositions placed on individuals’ behavior and romantic relationship, exclusively straight experiences. men were only exclusive for genital contact, and exclusively gay men were not exclusive on any indicator. Based on these findings, just as some gays/lesbians have nonpre- Funding ferred sex/romantic interests, so do some heterosexuals. Finally, in regard to bisexuality, it is difficult to dis- This research was funded in part by the American agree with Haslam (1997, p. 868) that bisexuality “may Institute of Bisexuality and the United States Department simply be a matter of people acting in accordance with of Agriculture’s Federal Formula Funds given to Cornell their continuously graded dispositions.” Construed in University. The funders had no role in study design, data this manner, the broader term bisexuality is less a cate- collection, analysis, decision to publish, or preparation of gory and more a continuous range of relative degrees of the manuscript. attraction, fantasy, infatuation, genital contact, and/or romance toward women and men—with variability across References individuals and indicators. In the data reported here, one bisexual person might be nonexclusive in fantasy and Austin, S. B., Conron, K. J., Patel, A., & Freedner, N. (2007). Making behavior, while another might be nonexclusive in infatua- sense of sexual orientation measures: Findings from a cognitive tion and romance; a bisexual might be 80% or 40% ori- processing study with adolescents on health survey questions. ented toward the opposite or the same sex with variability Journal of LGBT Health Research, 3, 55–65. dependent on the indicator. Thus, in addition to indicat- Bailey, J. M. (2009). What is sexual orientation and do women have one? In D. A. Hope (Ed.), Nebraska Symposium on Motivation: Vol. 54. ing nonexclusivity, bisexuality has further meaning when Contemporary perspectives on lesbian, gay, and bisexual identities it is specified in terms of its indicators. (pp. 43–63). New York, NY: Springer. The most important limitation of the present study is Baumeister, R. F. (2000). Gender differences in erotic plasticity: The that participants were not drawn from a representative female sex drive as socially flexible and responsive. Psychological sample and thus results are not necessarily generalizable Bulletin, 126, 347–374. to other populations. Participants were specifically Chandra, A., Mosher, W. D., & Copen, C. (2011, March). Sexual behavior, sexual attraction, and sexual identity in the United States: Data recruited for a study on sexuality and to enhance diver- from the 2006–2008 National Survey of Family Growth. National sity in sexual orientation. In particular, no general con- Health Statistics Reports, Number 36. U.S. Department of Health clusions should be reached regarding the prevalence of and Human Services. sexual orientation labels or a sexual spectrum in the larger Diamond, L. M. (2008). Sexual fluidity: Understanding women’s love and population or to other age groups. Neither is it known if desire. Cambridge, MA: Harvard University Press. participants had a shared understanding of the sexual Ellis, L., Robb, B., & Burke, D. (2005). Sexual orientation in United States and Canadian college students. Archives of Sexual Behavior, orientation indicators, not only in terms of definition but 34, 569–581. also in intensity, quality, and frequency. For example, a Friedman, M. S., Silvestre, A. J., Gold, M. A., Markovic, N., Savin- male in a sexual fantasy might be a peripheral figure or a Williams, R. C., Huggins, J., & Sell, R. L. (2004). Adolescents central component necessary to maximize the intensity define sexual orientation and suggest ways to measure it. Journal of of the fantasy. Genital contact includes not only touching Adolescence, 27, 303–317. someone’s genitals but also intercourse, with perhaps Gangestad, S. W., Bailey, J. M., & Martin, N. G. (2000). Taxometric analyses of sexual orientation and gender identity. Journal of considerably different motivations and meanings for indi- Personality and Social Psychology, 78, 1109–1121. viduals. Partners might differ in terms of whether their Haslam, N. (1997). Evidence that male sexual orientation is a matter relationship constitutes a romance. Whenever studies can of degree. Journal of Personality and Social Psychology, 73, afford detailed questions, more specific information 862–870. regarding the meaning, frequency, intensity, and quality Hoburg, R., Konik, J., Williams, M., & Crawford, M. (2004). Bisexuality of sexuality should be asked. In addition, given the nature among self-identified heterosexual college students. In R. C. Fox (Ed.), Current research on bisexuality (pp. 26–36). Binghamton, of data collection and the limited number of participants, NY: Harrington Park Press. potential differences related to race, ethnicity, and class Horovitz, B. (2012, May 3). After Gen X, Millennials, what should next were not addressed. generation be? USA Today. 452 NATURE OF SEXUAL ORIENTATION

Jones, R. P., & Cox, D. (2010). Two decades of polling on gay and lesbian Savin-Williams, R. C. (2006). Who’s gay? Does it matter? Current issues at Pew: An overview and assessment analysis. Washington, Directions in Psychological Science, 15, 40–44. DC: Public Religion Research Institute. Savin-Williams, R. C., Joyner, K., & Rieger, G. (2012). Prevalence and Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior stability of self-reported sexual orientation identity during young in the human male. Philadelphia, PA: Saunders. adulthood. Archives of Sexual Behavior, 41, 103–110. Klein, F., Sepekoff, B., & Wolf, T. J. (1985). Sexual orientation: A multi- Savin-Williams, R. C., Rieger, G., & Rosenthal, A. M. (2013). variable dynamic process. Journal of Homosexuality, 11, 35–49. Physiological evidence for a mostly heterosexual orientation LeVay, S., & Baldwin, J. (2012). Human sexuality (4th ed.). Sunderland, among men. Archives of Sexual Behavior, 42, 697–699. MA: Sinauer. Savin-Williams, R. C., & Vrangalova, Z. (2013). Mostly heterosexual as McConaghy, N. (1987). Heterosexuality/homosexuality: Dichotomy or a distinct sexual orientation group: A systematic review of the continuum. Archives of Sexual Behavior, 16, 411–424. empirical evidence. Developmental Review, 33, 58–88. McConaghy, N. (1999). Unresolved issues in scientific sexology. Archives Sell, R. L. (1997). Defining and measuring sexual orientation: A review. of Sexual Behavior, 28, 285–318. Archives of Sexual Behavior, 26, 643–658. Morales Knight, L. F., & Hope, D. A. (2012). Correlates of same-sex Taywaditep, K. J., & Stokes, J. P. (1998): Male bisexualities. Journal of attractions and behaviors among self-identified heterosexual uni- Psychology and Human Sexuality, 10, 15–41. versity students. Archives of Sexual Behavior, 41, 1199–1208. Vrangalova, Z., & Savin-Williams, R. C. (2010). Correlates of same-sex Peplau, L. A. (2001). Rethinking women’s sexual orientation: An inter- sexuality in heterosexually identified young adults. Journal of Sex disciplinary, relationship-focused approach. Personal Relationships, Research, 47, 92–102. 8, 1–19. Vrangalova, Z., & Savin-Williams, R. C. (2012). Mostly heterosexual Preciado, M. A., Johnson, K. L., & Peplau, L. A. (2013). The impact of and mostly gay/lesbian: Evidence for new sexual orientation identi- cues of stigma and support on self-perceived sexual orientation ties. Archives of Sexual Behavior, 41, 85–101. among heterosexually identified men and women. Journal of Weinberg, M. S., Williams, C. J., & Pryor, D. W. (1994). Dual attraction: Experimental Social Psychology, 49, 477–485. Understanding bisexuality. New York, NY: Oxford University Rodríguez Rust, P. C. R. (2002). Bisexuality: The state of the union. Press. Annual Review of Sex Research, 13, 180–240. Weinrich, J. D., & Klein, F. (2002). Bi-gay, bi-straight, and bi-bi: Savin-Williams, R. C. (2005). The new gay teenager. Cambridge, MA: Three bisexual subgroups identified using cluster analysis of the Klein Harvard University Press. Sexual Orientation Grid. Journal of Bisexuality, 2, 110–139.

453 Copyright of Journal of Sex Research is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.